AI's Promise and Peril: A County Perspective

That distinction matters most for the people counties serve every day: the family applying for SNAP, the senior seeking home care, the young adult aging out of foster care, the worker between jobs who needs temporary assistance and a path back to employment. These residents rarely have a seat at the table when technology decisions are made. Counties must make sure they are the first people considered.

The promise is genuine. At NYSAC's first AI Summit in July 2025, county leaders described tools already at work: chatbots that answer common social services and HR questions around the clock, scanners that pull data from handwritten applications, and software that summarizes long documents in minutes. Using a simple scoring rubric, participants identified high-value, lower-risk starting points such as AI-assisted review of 911 calls, warning systems at low-clearance bridges, and templates for county board actions.

Used well, these tools can give caseworkers back hours now lost to paperwork, time they can spend with the people in front of them. That is the right measure of success: not how many positions AI eliminates, but how much better residents are served. As Jefferson County's Dylan Soper put it at the Summit, humans must always be the first and the last step when using AI.

The perils are just as real. When an algorithm helps decide who qualifies for benefits, who gets flagged for review, or which case moves first, errors are not abstract. A wrong answer from a chatbot can mean a missed deadline, a lost benefit, or a family without heat in January. Systems trained on historical data can provide incorrect information or repeat historical bias. Pope Leo XIV's recent encyclical, Magnifica Humanitas, warns that the exclusion of the vulnerable can be cloaked "in a veneer of neutrality and objectivity." County leaders of every faith, or none, should take that warning seriously.

There is also the question of who benefits. New technology tends to reward those who already hold the data, the computing power, and the capital. Public data, built from the records of our residents, should not become private profit for a handful of companies while the public carries the risk. The New York Times editorial board recently argued that AI developers should answer in court for any harm their products cause, just as makers of cars and medicines do. Counties should insist on the same accountability in every contract they sign, and procurement, IT, and legal officials should be empowered to make that happen.

Counties will also soon face a very physical question: where the data centers go. These facilities consume enormous amounts of electricity and water. County and local officials will have to weigh the jobs and tax base they promise against strain on the power grid, local water supplies, farmland, ratepayers' bills, and noise levels. Those decisions belong in public view, with well-informed communities.

The answer is neither to reject AI nor to rush to embrace it, but to adopt it deliberately. NYSAC's 2025 report, Navigating AI in County Operations, lays out a roadmap: form a cross-departmental AI task force, adopt a clear use policy, start with small pilots, train staff, and measure results by time saved and residents served. Our GovAI Trustmark, developed by county leaders in partnership with NYSAC and Computer Aid, Inc., lets counties compare tools on shared criteria for cost, security, data protection, and risk, so no county has to size up a vendor alone.

Every county's AI journey will look different. But each should be guided by the same question: does this serve the common good of the most people, especially those who rely on us most? If the answer is yes, counties should move forward with confidence. If it is unclear, they should slow down and ask harder questions. That is not resistance to progress. It is what responsible public service has always demanded.